Attitudes were defined as “a psychological tendency that is expressed by evaluating a particular entity with some degree of favor or disfavor” (Eagly and Chaiken, 1993, p. 1). Teachers' attitudes toward inclusive education specifically refer to their beliefs, feelings, and intentions toward an inclusive policy. The scientific literature tends to show that these attitudes are somewhat positive and that the more they are, the more teachers' are willing to include students with SEN in mainstream education (Wertheim and Leyser, 2002; MacFarlane and Woolfson, 2013; Wilson et al., 2016; Hind et al., 2019), the more welcoming is the classroom learning environment (Monsen et al., 2014) and the more students with SEN feel socially integrated (Ben-Yehuda et al., 2010; Heyder et al., 2020). However, these attitudes are also ambiguous (Scruggs and Mastropieri, 1996; Krischler et al., 2018; for a review see van Steen and Wilson, 2020) since teachers generally support the philosophy of inclusive education but are reluctant to implement it in their own classrooms (Alghazo and Naggar Gaad, 2004; Hwang and Evans, 2011).

Many studies have been conducted to identify variables related to teachers' attitudes toward inclusive education. The most commonly targeted categories are teacher-related, student-related, and environment-related variables. Teacher-related ones refer to teachers' experience (Glaubman and Lifshitz, 2001; Alghazo and Naggar Gaad, 2004), gender (Alghazo and Naggar Gaad, 2004; Parasuram, 2006), training (Tournaki and Samuels, 2016; Lautenbach and Heyder, 2019), self-efficacy (Pit-ten Cate et al., 2018), values (Perrin et al., 2021), or position (McHatton and Parker, 2013; Desombre et al., 2019). Student-related variables identified as having an influence on teachers' attitudes include the type of disabilities/difficulties the students exhibit (Krischler and Pit-ten Cate, 2019; Jury et al., 2021a,b), and the extent to which instructional practices should be modified to meet students' special needs (Center and Ward, 1987). For example, teachers expect students with emotional and behavioral difficulties as well as those with profound and complex learning difficulties to be the most difficult to include (Avramidis and Norwich, 2002).

Finally, environment-related variables refer to the country's educational policies (Savolainen et al., 2012), cultural differences among teachers (Leyser et al., 1994), educators' understanding of inclusive education policy (Krischler and Pit-ten Cate, 2019), or school resources and support available for teachers (Minke et al., 1996; Avramidis and Norwich, 2002). To achieve inclusive education, teachers need to be confident both in their personal competences to meet the special needs of their students and in the collective competences to move toward a more inclusive school (Curchod-Ruedi et al., 2013). Collective competence refers to the support a teacher can receive in his or her classroom. According to Brackenreed (2008), one of the most important sources of stress for teachers who include in their classrooms students with SEN is the lack of adequate support, which gives rise to the feeling of being alone to deal with the situation. This may be true even for teachers who agree in general with the policy of inclusive education.

Social support has been the subject of growing scientific interest since the 1970s. Despite this expansion, this construct has been defined in many ways without reaching consensus among investigators (Pearson, 1986). Indeed, various specific types of social support have been identified with no agreement regarding neither their number nor types (Barling et al., 1988). At school, and for the present study, three types of social support can be identified (Langford et al., 1997). More precisely, teachers could receive instrumental support, which refers to providing assistance such as material goods and services or sharing of tasks or responsibilities, informational support, which relates to providing suggestions, advice, feedback, and problem-solving assistance, and emotional support referring to exchanges of empathy, expressions of caring, concern, affection, or encouragement.

Studies conducted in different contexts (professional, health, social networks) have shown that social support helps to reduce depression (Hays et al., 1990) and physical distress (Leserman et al., 1999) and bolsters psychological well-being (Hays et al., 1990). It should be noted that the concept of social support includes both perceived and actual social supports (McDowell and Serovich, 2007). Perceived social support appears to have more effect on health than actual availability of social supports (McDowell and Serovich, 2007).

Regarding inclusive education, social support can come from a variety of sources (Villa et al., 1996; Werts et al., 1996): peers, administrators, superiors, or resources (training, support from a team of experts, and support in the classroom). The link between these various sources of support and teachers' attitudes toward inclusive education has been identified in several studies. For example, peers support was identified as being an important influence on teachers' positive attitudes toward inclusive education policies (Boyle et al., 2012). In the same way, Rodríguez et al. (2012) showed that being part of a support network improved special educators' attitude toward the inclusion of students with autism spectrum disorders (ASD). Training has also been found to support teachers in the implementation of inclusive education and improve their attitudes toward inclusive education (Tournaki and Samuels, 2016; Pit-ten Cate et al., 2018; Li et al., 2019).

Although aimed at similar goals, the foregoing studies are intrinsically limited. Indeed, they either focus on specific SENs (i.e., students with ASD, Rodríguez et al., 2012; Li et al., 2019) or on specific teacher profiles (Tournaki and Samuels, 2016; Li et al., 2019). In addition, these studies focus on only one source of support, which suggests a need to study the role of social supports in a more general way. Finally, the methods used in these studies are mainly qualitative or correlational. Thus, using an experimental design would allow a sounder understanding of how social support predict teachers' attitudes toward inclusive education.

In the present study, we attempted to better understand the influence of social support on teachers' attitudes toward inclusive education. To do so, we first aimed to replicate in a pilot study the correlational link between these two constructs in the French context. Indeed, in addition to the importance of replication in educational and social psychology (Open Science Collaboration, 2017), the cultural differences in the inclusive education policy and corresponding teachers' attitudes (see for example, Moberg et al., 2020) imply that previous results from other countries could not be taken for granted. Then, we proposed to test the causal link between social support and teachers' attitudes within a controlled experiment. Based on the literature cited above (Minke et al., 1996; Avramidis and Norwich, 2002), we hypothesized that inducing a positive social support would enhance teachers' attitudes toward inclusive education.

In this study no data were excluded, all data were collected before any analyses were conducted, and all variables that were analyzed are reported. All material and data regarding this project can be accessed at: https://osf.io/6jgex/.

Two hundred and sixty-two teachers (163 general education teachers and 98 special educators, 1 missing value) participated in this pilot study. The sample included 203 females, 58 males and one who did not identify as either male or female. Mean age was 41.75 (SD = 9.73); mean teaching experience of 17.51 years (SD = 10.27).

During the spring semester of the 2018–2019 school year, potential volunteers were invited by email to participate in an online study. The email was sent to and by academic authorities, so the investigators did not know the potential participants. This email informed them about the study's main topic (i.e., inclusive education) as well as its procedure (i.e., an online questionnaire that should be answered in 10 min). In addition, they learned that their participation was voluntary, that they could quit the study at any time without consequences and that they would not receive any financial compensation. Upon consenting, participants were asked to fill in the questionnaires. At the end, they were debriefed regarding the goal of the study.

To test the hypothesis that the more teachers perceive that they receive social support, the more positive their attitudes will be, a regression analysis has been conducted. As anticipated, results confirmed that the more teachers perceived social support regarding the inclusion of students with SEN, the more positive their attitudes toward inclusive education, B = 0.23, t₍₂₆₀₎ = 8.38, p < 0.001, η_p² = 0.21, 95% CI [0.17; 0.28]¹.

Results from this pilot study confirmed the hypothesis in the French context. However, as mentioned above, the study's correlational nature limits the extent to which causal conclusions could be drawn regarding the impact of social support on teachers' attitudes. Consequently, a pre-registered experiment was conducted to test the hypothesis assuming that a positive social support could improve teachers' attitudes in comparison with a negative social support or a control condition in which social support would not be mentioned.

As indicated in the preregistered form (AsPredicted#34492¹), an a priori power analysis performed with G^*power 3.1 (Faul et al., 2007) revealed that 152 participants would be needed to detect a small effect (f = 0.22) with targeted power of 0.80. This was the minimum targeted threshold. Ultimately, 314 pre-service and in-service teachers participated in the data collection. The sample included 240 females, 72 males and two who did not identify as either male or female. For the whole sample, the mean age was 34.29 years (SD = 11.07) and the mean teaching experience was 8.97 years (SD = 10.66). Among the participants, 124 were general education teachers, 50 were special educators and 140 were pre-service general education teachers. No information about their experience with students with SEN was collected.

During the fall semester of the 2019–2020 school year, potential volunteers were invited by email to participate in an online study. As in the pilot study, this email informed them about the study's main topic (i.e., inclusive education) as well as its procedure (i.e., a 10 min online questionnaire). As previously, participants read that their participation was voluntary, that they could quit the study at any time without consequences and that they would not receive any financial compensation. Upon giving consent, participants were asked to read the instructions and fill in the attitudes measure. At the end, they were debriefed regarding the goal of the study.

As indicated in the pre-registration form, and due to the specificity of our hypothesis, we have decided to conduct contrast analyses (see Brauer and McClelland, 2005). Indeed, “precise conclusions can be obtained from contrast analysis because a contrast expresses a specific question about the pattern of results of an Anova” (see Abdi and Williams, 2010, p. 253). Therefore, the regression model tested two contrasts: the contrast of interest compared the positive support condition (coded +2) to the control condition and the negative support condition (coded −1 each), assuming that participants in the control condition perceived low support. The orthogonal contrast compared the control condition (coded +1) to the negative support condition (coded −1, the positive support condition being coded 0). To fully support our hypothesis, the contrast of interest should be significant while the orthogonal one should not. Preliminary analysis indicated that three participants had studentized residuals below −3.29 and thus were cause for concern and deleted from the present sample (Field et al., 2012)².The regression analysis revealed that the contrast of interest is significant, B = 0.07, t₍₃₁₁₎ = 2.81, p = 0.005, η_p² = 0.02, 95% CI [0.02; 0.11], indicating that participants in the positive support condition (M = 4.05, SD = 0.57) express more positive attitudes than participants in the control condition (M = 3.97, SD = 0.57) or the negative support condition (M = 3.73, SD = 0.66). However, it should be noted that the orthogonal contrast is also significant, B =0.12, t₍₃₁₁₎ = 2.92, p = 0.004, η_p² = 0.02, 95% CI [0.03; 0.20], indicating that the contrast of interest did not fully explain the variance and that participants in the negative support condition tend to express more negative attitudes than participants in the control condition. In other words, our hypothesis is only partially supported by our data.